Finite-Bandwidth Control of Fast Electrons Produced by Parametric Instability

In a microwave experiment simulating laser-plasma interactions, the production of fast electrons by the parametric decay instability is found to be significantly decreased by replacing a monochromatic pump with a noisy one. With this experimental arrangement, it is observed that finite bandwidth also inhibits the formation of density cavitons. We have experimentally presented the effects of finite-bandwidth pumps on the parametric instability at power levels sufficient to produce hot electrons.

The experiments were performed in an unmagnetized electrostatically confined filament-discharge plasma of 100 cm length and 35 cm diameter with argon fill pressures of 0.4 mTorr.

The plasma density as a function of the distance from the dielectric window located at the narrow (entrance) end of the microwave horn is displayed in the inset of Fig. 1. The mean density of the pulsed plasma changes by less than 0.1% over the rf pulse duration. The corresponding change in plasma frequency is much less than the observed instability resonance width.


Fig. 1. Hot-electron current (|Ue|>125 eV) as a function of pump bandwidth for both noise phase-modulated pumps, and noise amplitude-modulated pumps. Inset shows the plasma density profile.

At moderate rf power levels (P>7 W and w0~wp) the low-frequency density fluctuations and high-frequency electric field fluctuations characteristic of the parametric decay instability are observed. The instability occurs within the uniform density region located 10-15 cm from the input window (see inset Fig.1). Upon instability onset, as detected by shielded Langmuir probes of both the double and single types, an increase in the hot-electron current to the analyzers is observed. The energetic electrons are found to be preferentially directed along the electric field of the incident microwave radiation. Varying the frequency of the narrow-band pump, we found that the resonance width for hot-electron production increases from 2.4% of the pump frequency near threshold power to 5% at higher powers (P~300 W). The effect that the pump bandwidth can have a significant effect on hot-electron production when pump bandwidth is larger than resonance bandwidth was investigated in the present experiment using both noise-phase-modulated pumps with bandwidths variable up to 1.2% and noise-amplitude-modulated pumps with bandwidths up to 23%.


Fig. 2. Hot-electron current as a function of pump center frequency for a narrow-band pump o and a noise amplitude-modulated pump . Pump power is 340 W for (a), 142 W for (b), 51W for (c), and 31 W for (d).

The saturated hot-electron flux is shown in Fig. 2 for both the narrow-band and 300 MHz (Dw/w0 ~10%) noise-amplitude-modulated pumps as a function of pump center frequency for several power levels. Near the instability resonance center frequency, the finite bandwidth pump is observed to reduce or even eliminate the analyzer hot-electron current. However, at high power, the effective instability resonance width is seen to increase for the noise pump. In the case of the broadband pump, hot electrons are detected at pump center frequencies where non are observed with the narrow pump. This phenomenon may partially explain the increase in the instability amplitude observed by Yamanaka et al. in a finite-bandwidth laser heating experiment.

The effect of finite bandwidth on the instability-produced hot-electron distribution function was investigated. For our plasma, the high-power narrow-band pumps heat a fraction (<10%) of the electrons in the 20 eV tail of the initial electron distribution function to a higher-temperature (~30 eV) Maxwellian velocity distribution. Similarly, we found that the finite-bandwidth pumps simply decreased the maximum number of heated electrons, without changing the temperature of the heated electrons. The relatively small percentage of heated electrons (<0.1% of the total) may reflect the small size of the instability region in comparison to the hot-electron mean free paths.


Fig. 3. Hot-electron current as a function of time. (a) Narrow-band pump. (b) Noise amplitude-modulated pump.

Fig. 3 displays the time evolution of the hot-electron current detected by an energy analyzer for both the narrow and noise-modulated (Dw/w0~10%) pumps at several power levels. At low power levels (P<~21 W ~ 3 Pth), there were no detectable hot electrons for the wide-band pump, while electron tail heating was observed down to 7 W with the narrow pump. At higher powers, P>~21 W) hot electrons are produced by both pumps and we can compare their respective rates for hot-electron production. The initial fast growth in the hot-electron current which saturates and begins to decay ~ 1 usec after turn-on for the narrow pump [Fig. 3(a)] may be due to cavitation or resonance absorption arising from the plasma density gradient along the pump electric field. Finite pump bandwidth appears to drastically inhibit this early quickly saturated hot-electron flux produced by the wide-band pump during the first few microseconds is comparable to or larger than that of the narrow pump. When the pump center frequency is not adjusted to narrow-band threshold minimum, the wide-band pumps can have consistently faster growth rates for a given power than the narrow-band pumps, similar to the saturation results of Fig. 2.


Fig. 4. Temporal evolution of electron density perturbations produced by 340 W short-duration rf pulses. (a) 100 nsec narrow-band rf pulse, (b) 400 nsec narrow-band rf pulse, (c) 400 nsec random amplitude-modulated rf pulse (Dw/w~10%).

The plasma density disturbances associated with the initial fast-electron production mentioned above were investigated using short rf pulses of variable duration (20-500 nsec). In Fig. 4(a), we show the electron density perturbation, as detected by a Langmuir probe biased to electron saturation, which occurred upon irradiating the plasma with a 100 nsec narrow-band rf pulse at high power. Similar density perturbations, but with reduced amplitude, were observed for rf pulses as short as 40 nsec. Fig. 4(b) shows the result of employing a longer narrow-band pulse with the same average power, pulse duration, and center frequency as the narrow-band pump in Fig. 4(b). The bursts of hot electrons which were observed to occur along the pump field with the short narrow-band rf pulses were also nearly eliminated with the wide-band rf.

In conclusion, we find that finite pump bandwidth can significantly increase the minimum threshold and reduce the saturation level of parametric decay-produced suprathermal electrons. It also can apparently control the hot-electron production due to cavitation. The fact that finite bandwidth did not produce a marked decrease in heating rates for nonthermal electrons on all time scales is somewhat surprising.